Increasing intracellular dNTP levels improves prime editing efficiency

Prime editing is a ‘search-and-replace’ genome editing technology that directly ‘writes’ new genetic information into a targeted DNA site (Anzalone et al.). As discussed in the next section, prime editing utilizes an engineered reverse transcriptase for this writing, which incorporates endogenous deoxynucleotide triphosphates (dNTPs) guided by a synthetic RNA template. Consequently, the available pool of dNTPs can influence the overall efficiency of prime editing.??

This blog outlines how a team of researchers at the University of Massachusetts has been able to increase the level of dNTPs in order to improve prime editing efficiency. Their publication (Liu et al.) in Nature Communications on September 25, 2024, which has already received more than 3,700 accesses, uses a functional variety of in vitro transcribed (IVT) mRNAs each synthesized with TriLink’s CleanCap? cap analog and N1-methylpseudouridine-5'-triphosphate (m1Ψ-5’-TP).?

Prime editing mechanism?

As depicted in Figure 1’s steps 1-5 and discussed in detail elsewhere (Chen et al.), prime editing uses two main components: an engineered Moloney murine leukemia virus reverse transcriptase (MMLV-RT) fused to a Cas9 and a prime editing guide RNA (pegRNA). After binding of the pegRNA guide sequence (blue) to a complementary strand of DNA (1), the displaced “PAM” strand of target DNA is cut (“nicked”) by the Cas9 nickase. This cut exposes a DNA 3’-hydroxyl group in DNA to prime synthesis upon hybridization (2) with the primer-binding site of pegRNA, i.e., (3) reverse transcription?of the RT template portion of the pegRNA that encodes the desired edit (gold/red). ?

(4) Hybridization of the resultant DNA strands results in either a 3’ flap that contains the newly synthesized sequence, which can be cleaved (edit is removed), or a 5’ flap that contains the unedited DNA sequence, which can be cleaved (enables editing). (5) Subsequent 3’ flap ligation creates a mismatched DNA “bubble” for repair with two possible outcomes: the information in the edited strand is copied into the complementary strand (edit is incorporated); or the original nucleotides are re-incorporated into the edited strand (edit is removed). ?

A number of prime editing modifications were made to the original version (PE1) leading to an evolved series (PE2, PE3, etc.) culminating in ‘‘PEmax’’ prime editor architecture, which involved combined optimization of previous versions and other considerations (Chen et al.). Despite these advances, Liu et al. note that precise editing rates remained modest in some primary cell types and in vivo in some tissues. They hypothesized that prime editing efficiency in cells is hampered by inherent properties of the MMLV-RT.???

For example, most prior optimization of prime editing components has been performed in rapidly dividing mammalian cells; however, because intracellular dNTP concentrations can be 100-fold lower in postmitotic or quiescent cells than in cycling transformed cell lines, this scarcity of dNTPs could suppress MMLV-RT processivity due to this enzyme’s low affinity for dNTPs.?

To test this hypothesis with an aim to find methodology to improve prime editing rates, Liu et al. investigated the two-pronged strategy outlined below, namely, finding mutant MMLV-RTs exhibiting improved processivity by binding dNTPs better, and increasing the intracellular concentration of available dNTPs.?

Mutant MMLV-RTs for improving prime editing efficiency?

Optimization of the activity of MMLV-RT was investigated by screening several previously described MMLV-RT variants. Specifically, MMLV-RT processivity under limiting dNTP levels is enhanced by the Q221R mutation and various mutations at position 223 (Operario et al.), while MMLV-RT solubility is improved by an L435K mutation, without substantial impact on its catalytic performance (Das and Georgiadis).??

Liu et al. therefore used site-directed mutagenesis to evaluate the impact of Q221R and three different V223 mutations (V223A, V223M and V223Y) as well as L435K, individually or in combination, on prime editing rates to discern differences in activity. Preliminary comparisons of edits in various cell types indicated that relative to the MMLV-TR in PEmax as a benchmark, the single-mutant (*) V223Y variant was superior, and the double-mutant (**) L435K/V223Y was the most efficient. Prime editing with these single- and double-mutants is termed PEmax* and PEmax**, respectively. ?

Relative efficiencies of PEmax, PEmax*, and PEmax** were then studied using two independent editing methods: (i) synthesis of the MMLV-RT proteins for editing as ribonucleoprotein (RNP) complexes; and (ii) editing by expression of the MMLV-RTs encoded in CleanCap? m1Ψ-modified mRNAs. Each editing method was independently applied to editing the previously studied (Anzalone et al.) HEK4 and FANCF sites in the human genome.??

Briefly, the editing efficiencies obtained with mRNAs outperformed those obtained with RNP complexes for HEK4 and FANF in human skeletal muscle stem cells. Moreover, the found order of CleanCap? m1Ψ-modified mRNA prime editing efficiency is PEmax < PEmx* < PEmax**.??

That this superior utility of PEmax** can be generalized was supported by demonstrating that PEmax** efficiently edited genomic FANCF in human primary CD4+ T cells. In addition, PEmax** introduced a Q155H mutation into Pcsk9 in mouse liver by delivering lipid nanoparticles (LNPs) containing PEmax** or PEmax mRNA along with pegRNA and a “helper” nicking gRNA, as described previously (Davis et al.). Following retro-orbital injection, 1.8% precise editing at Pcsk9 in mouse liver was observed with PEmax**, a 1.7-fold improvement over PEmax.?

Increasing dNTP concentration for improved prime editing efficiency?

The results of the above in vitro and in vivo prime editing studies were consistent with intracellular dNTP levels representing a bottleneck for MMLV-RT-based prime editing. Liu et al. recognized that SAMHD1 is a dNTP hydrolase (Figure 2) that maintains low dNTP levels in cells that are not undergoing DNA replication, which restricts infection by retroviruses (Coggins et al.). Furthermore, the HIV-2 accessory protein VPX degrades SAMHD1 to increase intracellular dNTP levels, which promotes reverse transcription and successful viral infection in quiescent cells (Hofmann et al.).??

Therefore, Liu et al. investigated prime editing rates in HEK293T cells, U2OS cells, and primary fibroblasts for the impact of (i) co-delivering CleanCap? m1Ψ-modified mRNAs each encoding VPX or a PEmax variant (“VPX co-delivery”); or (ii) supplementing cellular growth medium with membrane-transportable deoxynucleosides (dNs). Because excess dNs can negatively impact cell fitness and proliferation, it was necessary to first evaluate the impact of dN concentration on cell viability to find a dN concentration affording adequate cell fitness.?

It was observed that VPX co-delivery or dN supplementation enhanced prime editing rates. VPX in combination with PEmax** increased prime editing rates ~3.5-fold relative to PEmax alone. Interestingly, dN supplementation in combination with VPX co-delivery produced similar increases in prime editing efficiency as observed with either approach in isolation, i.e., there was no apparent synergy.?

Consistent with intracellular levels of dNTPs playing a critical role in precise editing outcomes, as determined by sequencing, co-delivery of CleanCap? m1Ψ-modified mRNA encoding SAMHD1 suppressed precise editing rates and increased imprecise editing outcomes.?

Additional experiments demonstrated that VPX co-delivery increased precise editing rates for PEmax, PEmax* and PEmax** at the FANCF locus in resting T cells and significantly increased prime editing rates at the CCR5 locus in activated T cells. Similarly, VPX co-delivery significantly improved prime editing efficiency with PEmax** in stem cell-derived islet cells.?

?

Concluding comments?

The above results, and those from many additional experiments not mentioned in this blog, established that the activity of MMLV-RT-based prime editing can be improved by the use of IVT mRNA encoding mutant MMLV-RTs having increased dNTP affinity and solubility. Notably, the double-mutant PEmax** editor outperforms the previously optimized PEmax benchmark by ~1.5- to 2.5-fold across a range of target sites and cell types.?

Increasing intracellular dNTP levels through VPX co-delivery or supplementing dNs increases prime editing rates for all PEs that were tested and increases the ratio of precise edits to undesired edits, thus providing two advantages, namely efficiency and precision.?

Liu et al. envision that VPX co-delivery or dN supplementation will increase precise editing rates for a variety of editing applications. For editing cultured cells, both VPX co-delivery and dN supplementation produce similar enhancements in editing outcomes. However, they prefer VPX co-delivery when feasible, as cellular toxicity from transient SAMHD1 knockdown was not observed, whereas dN supplementation can lead to cellular toxicity if not properly controlled.??

In addition, VPX co-delivery, unlike dN supplementation, could potentially be used with in vivo delivery platforms (e.g., LNPs or engineered virus-like particle) to improve prime editing outcomes in organ systems with low intracellular dNTP levels.??

Given the growing interest in prime editing for basic R&D and clinical development, as reflected by the number of annual publications for 2019 to present in PubMed, the findings by Liu et al. will likely prompt further research using? IVT mRNAs encoding PEmax** and VPX.?

?

Your comments are welcomed, as usual.?